From Safety to Action to Impact: the summit arc

The summit series, 2023–2028

• 2025-02-01 — Frontier safety frameworks become table stakes:

  Following the Seoul commitments, major labs publish or update frontier safety policies (capability thresholds, evaluation gates, deployment mitigations) ahead of the Paris summit.

• 2023-11-01 — Bletchley Park AI Safety Summit:

  28 countries + the EU — including the US and China — sign the Bletchley Declaration on frontier-AI risk. The summit series begins.

• 2024-05-21 — Seoul AI Summit:

  Frontier labs sign safety commitments — publish risk frameworks or explain why not. The summit series turns from declarations to developer promises.

• 2025-02-10 — Paris AI Action Summit:

  The series pivots from safety to action and investment; the US and UK decline to sign the final declaration — the divergence made visible.

• 2026-02-19 — AI Impact Summit, New Delhi:

  The summit series lands in the Global South, centering development and inclusion; Geneva planned as the next stop (2027).

• 2026-07-06 — First UN Global Dialogue on AI Governance:

  The Global Digital Compact’s forum convenes in Geneva — every state at one AI governance table for the first time.

Bletchley Park, November 2023 — Safety

The UK convened 28 countries plus the EU — including China, a diplomatic coup nobody expected to survive later summits. The Bletchley Declaration named frontier-model risk, up to and including catastrophic harm, as a shared concern and committed signatories to state-backed testing and a scientific report process. The UK’s AI Safety Institute launched around it. Peak safety framing: the word ‘existential’ was nearly on the table, and the US and China signed the same page.

Seoul, May 2024 — Safety institutionalised

Co-hosted by South Korea and the UK. Three deliverables that outlived the communiqué: the Seoul Declaration; the Frontier AI Safety Commitments, in which 16 companies (later more) promised to publish frontier-risk frameworks with red lines and to refrain from deploying models breaching them; and the launch of the International Network of AI Safety Institutes. Seoul is the summit that converted rhetoric into standing machinery — which is why the capstone treats it as the arc’s hinge.

Paris, February 2025 — Action

President Macron’s rebrand: safety out of the title, investment in. Sixty-two signatories endorsed a declaration on ‘inclusive and sustainable’ AI — and the US and UK both refused to sign, the US objecting to the framing broadly, the UK citing insufficient attention to security. €100B-scale investment announcements, the Current AI public-interest foundation, and a pointed vice-presidential speech warning Europe against over-regulation. Paris marks the pivot: the safety coalition of Bletchley fractured on camera.

New Delhi, February 2026 — Impact

The first Global South host. India framed the summit around three ‘Sutras’ — People, Planet, Progress — with 100+ delegations and 20+ heads of state, and centred access, development and capacity rather than frontier risk. The second International AI Safety Report (Bengio-chaired, ~100 experts, 30+ countries) landed just before it — with the US withholding support. Critics called Delhi a trade fair with a declaration attached; defenders called it the first summit most of the world was actually invited to shape. Next: Geneva 2027, UAE 2028.